We report results from re-analysis of the visibility data of the first 2-beam observations with VERA (VLBI Exploration of Radio Astrometry), previously reported by Honma et al., 2003 (hereafter A2003). Independently we checked the archival data and found the features noted in A2003 were not from the effect of phase referencing by simultaneous differential VLBI but mainly from a removal of large phase change by subtracting an arbitrary fitted curve to the phase variations. The differential phase of the observed H2O masers between W49 North (W49N) and OH~43.8-0.1 did not show a sinusoidal variation with a period of one sidereal day due to a positional offset from the real celestial positions. We therefore could not reproduce the results in A2003 by a normal positional correction estimated from all time data, but could reproduce almost the same phases only for the first hour by adjusting parameters. Using the parameters, we could not suppress the large amount of phase variations for the successive time data that A2003 did not show in their paper. It is appropriate to regard the analysis in A2003 as not being proper for showing the performance of the instrument for phase referencing, which should be demonstrated by other experiments observing several pairs of continuum sources.